AIYIMA A07 TPA3255 Amplifier Measurements and Review - LM4562 (and OPA2134) option

[Deleted member 9685]

You better spend your money on a tube preamplifier because people say tube changes sound.

If LM4562 reduces noise leaking into house wires significantly, I may consider it a worthy investment. I like reducing noise in house wires.

 

[pma]

It does not. It slightly reduces amp measurable output noise, that’s all. The difference is not audible on music.

 

[DualTriode]

It does not. It slightly reduces amp measurable output noise, that’s all. The difference is not audible on music.

We are speaking of measurable things, everything adds up, everything counts. You are going to have a hard time convincing people what is not auditable and what is not important.

If you are going after after auditable with your measurement tools I would go after the cheapest things that will improve SNR and SINAD.

Thanks DT

 

[pma]

You are going to have a hard time convincing people what is not auditable and what is not important.

I have no need to convince anybody about audible or not. It has been a long path important for myself to find out what is audible and what is not. And I think I have found some definable boundaries, at least for myself. I am neither audiophile believer nor techno believer. I do not care much what the others do think about it.

 

[DualTriode]

pma ,

You tell us in one post that with music the noise is not audible. Then in the next post you tell us that you have no need to convince us if something is audible or not. Are you trying to own both sides of the argument?

What I care about is hearing the noise when the music stops.

Thanks DT

 

[spotifyguru]

All measurements are at 4ohms, how many watts can this supply single channel at 8 ohms before distorting? I'll probably run a 36v power supply.

 

[pma]

What I care about is hearing the noise when the music stops.

1.5dB improvement in background noise when NE5532 is changed to LM4562. Will you hear the difference even with your ear directly at the speaker?? -74.9dBV vs. -73.4dBV (24kHz measuring BW). Please answer to yourself.
I am not interested in endless debates about unimportant pseudo-issues. This amp would need a re-design of the input stage to get considerably lower noise level. That's all and that's simple. Audiophile juggling with opamps makes not much sense in this case.

 

[pma]

All measurements are at 4ohms, how many watts can this supply single channel at 8 ohms before distorting? I'll probably run a 36v power supply.

This is how to calculate it in case of ideal amplifier (no thermal shutdown) and power supply is not limited in supplied current :

Your power supply is 36Vdc, let's assume it is not limited in supplied current, i.e. it would give as much current as the amp needs. The equation for maximum power is then as in the image above. With your Vdc = 36V and assumed transistor saturation voltage Vces = 1V, you get 72W with 8ohm load and 144W with 4ohm load. You will need at least 3A to get 72W/8ohm and at least 6A to get 144W/4ohm.

 

[abdo123]

You will need at least 3A to get 72W/8ohm

can you share where that 3A figure comes from? or how different the situation would be if Vdc= 48?

 

[pma]

can you share where that 3A figure comes from? or how different the situation would be if Vdc= 48?

You can see the supply current as rms(i(V1)) in the plot above. It almost overlaps load current rms(i(R1)). Instant current is i(V1). As a simple equation, P = R1 * I(R1)^2. From that I(R1) = 3A. We are talking in rms terms of AC voltage and current.

For Vdc = 48V, Pmax = 132W/8ohm and 264W/4ohm. You need 4A or 8A. AIYIMA will never give such output power without shutdown, but it is just what TPA3255 is able of if properly used.

 

[Head_Unit]

Huh, I thought I had posted already: bought one of these to drive a set of passive JBL ControlX in a big room (loud but not earsplitting). It is essentially running as a power amp from an Airport Express, dead quiet with those speakers. Nice!

 

[friarminor]

After reading, changed OpAmp to LM4562 and bought 48V SMPS from Aiyima - thanks to Youtube didn't burn this one and was able to connect & adjust to 44V. So far, new opamp seem crispier with better instrument separation but new power supply definitely gave it louder oomph allowing me to hear those sound whispers in my KEFLS50s (which are proven to be hard to drive). But then again, I may also be fooling myself.

 

[Bob from Florida]

You can see the supply current as rms(i(V1)) in the plot above. It almost overlaps load current rms(i(R1)). Instant current is i(V1). As a simple equation, P = R1 * I(R1)^2. From that I(R1) = 3A. We are talking in rms terms of AC voltage and current.

For Vdc = 48V, Pmax = 132W/8ohm and 264W/4ohm. You need 4A or 8A. AIYIMA will never give such output power without shutdown, but it is just what TPA3255 is able of if properly used.

PMA - great review with extensive testing. The power supply is always the key ingredient and most expensive part of any given amp if seems. My question is have you tried using batteries? I understand there are limitations with that idea - charging challenges, etc. Seems like 3 - 12 volt deep cycle batteries in series for 36 volts could be interesting to try.

 

[pma]

PMA - great review with extensive testing. The power supply is always the key ingredient and most expensive part of any given amp if seems. My question is have you tried using batteries? I understand there are limitations with that idea - charging challenges, etc. Seems like 3 - 12 volt deep cycle batteries in series for 36 volts could be interesting to try.

No, I have not tried batteries. But I think that there is no need to do so. Because either with the regulated linear power supply or with SMPS we have no traces of supply mains residuals. The only concepts to avoid are linear unregulated power supply or cheap SMPS that produce high mains line residuals as a result of poor PSRR of the A07.

and the next one with Meanwell SMPS

 

[Bob from Florida]

No, I have not tried batteries. But I think that there is no need to do so. Because either with the regulated linear power supply or with SMPS we have no traces of supply mains residuals. The only concepts to avoid are linear unregulated power supply or cheap SMPS that produce high mains line residuals as a result of poor PSRR of the A07.

View attachment 155013


and the next one with Meanwell SMPS

View attachment 155014

I agree a good quiet switcher can work well. I have had experience with cheap switchers with 3 wire power cords injecting audible "swishing" sounds via the ground wire into the stereo system. A cheater plug confirmed the switcher was the source so I ended up using a Pyramid linear power supply long term.

 

[pma]

Yesterday I posted IMD measurements with a small systematic error, so here is a repeated version. Together with THD and THD+N at 1kHz. All into 8ohm load. Power supply is SMPS 27V. Please note that 1kHz THD+N (SINAD) gives misleading and too optimistic results re amplifier distortion behaviour. This would be my repeated request to @amirm to give more attention to IMD measurements and not to use this SINAD 1kHz parameter as a general quality measure in chart comparisons. It is not straightforward for frequency dependent and non smooth nonlinearity to guess IMD issues from 1kHz THD. It is only easy for polynomial Y = An*X^n + An-1*X^n-1 + ... + 1 simple nonlinearity. We can also see that intermodulation behaviour of this amplifier is really not good. It is much worse than for comparable well designed class AB amplifier. Together with FR heavily depending on load impedance, this might be the reason why many members like this "different sounding" amplifier.

I won't stop saying that this amplifier (and similar cheap class D amps) is doing a wrong job in signal amplification. The only usable class D amplifiers so far are NCore and Purifi based ones.

 

[pma]

Measurement of THD vs. frequency at 30W/4ohm. There are 2 plots, green has measuring BW=22kHz, red has BW=45kHz

Above 200Hz the harmonic distortion rises with 20dB/decade slope.

And measurements of THD 1kHz, THD 15kHz, IMD 60Hz+7kHz, IMD 250Hz+8kHz, IMD 19+20kHz vs. output voltage. Only THD 1kHz is respectable, other measurements show poor and unpredictable nonlinearity and especially very poor high frequency nonlinearity.

 

[pma]

I was owing a distortion measurement into 8 ohm load. So here it goes, at 1kHz, 8ohm and with 27Vdc SMPS power supply. Good up to 30W, then fast rise of noise background. Some 40W at THD+N 1%.

And, again to the topic of frequency response dependence on load impedance, below is the measurement of frequency response into 4ohm, 8ohm and my complex speaker dummy load. This the weakest point of all class D amplifier with output LC filter outside the feedback loop. No RC low pass filter to restrict measuring input BW was used, as it would affect the shapes above 10kHz.

And as a reminder, the dummy load impedance plot

 

[UAN]

From the frequency response of the load resistance, it can be seen that the amplifier has a significant output impedance. And what is it like at a frequency of 100 Hz?

 

[pma]

From the frequency response of the load resistance, it can be seen that the amplifier has a significant output impedance. And what is it like at a frequency of 100 Hz?

I think you understand that the output impedance is dictated by the LC output filter. In case of A07 LC output impedance can be seen below and it is same as circuits recommended in TI application notes. This is again no surprise, only that a common user probably not always realizes this fact.
2 plots are shown, with output filter loaded with 4ohm and 8ohm resistors. Without resistors, there is a high LC resonance peak.

Please take into account that the output impedance is inductive up to some 10kHz, it is not a resistor!!



Simulated freq response of the output filter to 4ohm and 8ohm load

From Bruno Putzeys AES paper, comparison of output impedance of class D UcD amplifier with LC filter outside and inside feedback loop:
(please note this is not the same as TPA3255 with PFFB, TPA3255 with PFFB cannot reach the response as shown in blue plot)